Skip to content
Projects

toki-plus/ai-mixed-cut has 94 GitHub stars and 10 forks. GitHot measured 0 stars in 24 hours, 0 in 7 days, and +1 in 15 days. Last checked 2026-11-17 03:30 UTC. Last push 2026-08-08.

How these counts are measured

Measured star growth

in 24h
0
in 7 days
0
in 15 days
+1
in 1 month
Collecting history

Net star changes over overlapping periods, not daily totals. Missing measurements are left blank. Stars measure attention, not software quality.

About

颠覆性AI内容再创作引擎,通过“解构-重构”爆款视频模式,全自动生成高度原创短视频。AIGC, Content Creation, Video Generation, Automation, LLM, Python, FFmpeg.

AI

README_en.md

Original

AI Mixed Cut

简体中文 | English

An AI-assisted video-workflow case study for authorized content repurposing.

The solution follows a material-analysis, structuring, script recomposition, speech-generation, and video-composition process. It explores how content teams can improve reuse of existing assets while keeping human review in control.

This repository is a curated public case study. It contains the solution description and interface examples, but not the complete implementation. All source material must be legally authorized for use and adaptation.

Problem Context

Content teams often accumulate large volumes of video, copy, and audio without a consistent structure. Finding relevant segments, organizing themes, drafting new scripts, and assembling outputs requires repetitive work across several tools.

The solution separates the workflow into independently verifiable modules with explicit inputs and outputs.

Intended Users

  • Content operations teams managing large historical media libraries
  • Editors breaking long-form content into focused topics
  • Product and technical teams validating AI-assisted content workflows

Solution Flow

Authorized source material
    -> Content extraction
    -> Structure analysis
    -> Material library
    -> Script recomposition
    -> Speech generation
    -> Video composition
    -> Human review

Main Modules

1. Content Analysis and Structured Material Library

  • Extract editable text and timeline information while preserving source references for traceability and review.
  • Deconstruct video content along three dimensions into a reusable, structured material library:
    • Topic decomposition: distills core subjects, target audiences, and value propositions into topics.json;
    • Framework decomposition: abstracts narrative structures and content formulas into frameworks.json;
    • Sentence decomposition: curates high-quality, broadly applicable sentences into golden_sentences.json.
  • The library is persisted as JSON files that later modules can read independently, and that editors can curate and version.

2. Script Recomposition and Audio Generation

  • Combine "topic + framework + sentence" entries from the library and generate candidate scripts with a model under explicit content constraints; titles and tags are drafts only and require human approval before use.
  • Integrate Edge TTS for multilingual voices with adjustable speed, pitch, and volume; timeline-aligned .srt subtitles are produced alongside the audio.
  • Subtitles go through a second-pass segmentation step that optimizes line breaks by semantics and layout. All intermediate files are preserved so editors can revise and re-run downstream steps.

3. Video Composition

  • Manage multiple source folders by topic; assemble clips sequentially or randomly, with per-clip fixed, randomized, or original durations.
  • Optional visual treatments — LUT color grading, dynamic zoom, random rotation, background blur, mask overlays, and xfade transitions — support differentiated re-creation of source material.
  • NVIDIA (NVENC) hardware encoding accelerates rendering; parameterized rendering and batch-task entry points are provided, and each module can run independently for debugging and verification.

Technical Considerations

  • Treat AI outputs as candidates, not publish-ready material
  • Preserve source, authorization, and processing records
  • Separate browser automation, model calls, and media processing into clear boundaries
  • Plan for external-service failures, timeouts, and interface changes

📸 Screenshots

Main application interface
Main interface: the modular workflow — content analysis, script recomposition, audio generation, and video composition — is visible at a glance.

Project Status

This repository presents the workflow design and product prototype. The complete implementation is not public because it contains third-party configuration and non-public business logic.

Responsible Use

  • Process only text, audio, and video with appropriate authorization.
  • AI-generated content requires factual, copyright, and brand review.
  • Users must comply with platform rules, provider terms, and applicable laws.

📂 More Projects

  • video-mover — Automated multi-platform content distribution pipeline: media processing, metadata generation, scheduling, platform adapters
  • ai-highlight-clip — Long-video smart triage: Whisper transcription + LLM scoring + human review
  • ai-ttv-workflow — Desktop text-to-video workflow with human-in-the-loop checkpoints
  • ai-video-workflow — Multi-model AIGC video pipeline orchestrating image, video and music services
  • ai-trader-for-mt4 — LLM×MT4 controlled-execution framework: constrained tools, risk rules, state management
  • ai-trader-for-mt5 — AI trading assistant and EA engineering framework for MetaTrader 5
  • auto-usps-tracker — Batch shipment tracking and Excel reporting for cross-border e-commerce
  • AB-Video-Deduplicator — Experimental video re-creation tool based on high-frame-rate blending
  • netease-downloader — Netease Cloud Music desktop downloader: QR login, queue, ID3 tagging

License

See LICENSE.